May 05, 2010
I just took the political compass test to see where I stand.
No surprise, I was -2.50 on the economic scale and -5.28 on the social authoritarian/libertarian scale. The test posed some interesting questions, but all in all I felt that there was a severe flaw. Namely, the questions were asked in a way that appealed to how we perceive an ideal world, rather than what is practically viable, given that we are human and must create a society that allows for normal human error.
Letâ€™s look at a specific example, and then attempt to generalize (though not rigorously by any means). Here was a question, which may not be exactly as written but presents the central theorem uncorrupted:
â€œOur rights are being impinged upon in the name of counter-terrorismâ€
Letâ€™s not even bother with the fact that the answer to this question has to be yes. If you have ever been on an airplane, you have sacrificed some rights, plain and simple.
But there is a more subtle flaw here. Letâ€™s suppose that the response to this binary question is instead given by a point, A, such that 0 <= A <=1. We interpret A=0 as believing that no rights should be sacrificed for counter-terrorism and A=1 being a belief that derogation of all rights is acceptable in the name of counter-terrorism. In practicality, we may allow for slightly more rights to be derogated. We let â€œLâ€ denote our acceptable limit in a practical reality. Then A <= L. In other words, our practical belief is limited by the answer to the question.
The test uses an algorithm to assess your answers to various questions and posts a final score as a point (x,y) in a 2-D box. Here, the x-axis represents economic beliefs and the y-axis representing social beliefs. If we assume that we have an ideal political bias and that this is carried through all questions, then we are free to generalize the above argument to all questions on the test. It would then seem reasonable to think that our ideal political inclination, (xâ€™,yâ€™), is bounded in some manner by the output (x,y).
Neutrality Proposition: It is my belief (here is the non-rigorous argument) that |x|-|xâ€™|>=0 and |y|-|yâ€™|>=0. What this means is that, once error is introduced into a perfect world, our beliefs tend toward neutrality. This can be interpreted as the statement: A liberal (resp. conservative) will get more conservative (resp. liberal) once human error is introduced into an ideal setting.
Now, we implicitly assume that the error is small relative to a given personâ€™s ideals, or else the inequality could fail due to a massive swing through the origin. For example take x = 1 and xâ€™=-5. Then the error is so large that the inequality does not hold. In reality, I donâ€™t think the world is harsh enough that it turns idealistic conservatives into strong liberals and vice versa. Hence I ignore this possibility.
Finally, if we accept the above proposition, then we see that |(xâ€™,yâ€™)| <= |(x,y)|. In other words, the test gives us a limit on our liberal or conservative viewpoints. So when Karen said she was surprised at the outcome of my test, I think she was noticing a disparity between my output (limit on my liberal inclination) with my practical beliefs, given real-world complications. The above argument just tries to rationalize this disparity.
There are clear holes in this argument. Obviously, the Neutrality Proposition is difficult to prove and may not even be correct. Furthermore, we are assuming that every answer is either <= 0.5 or >=0.5 (some sort of pseudo-monotonic assumption). Despite these gaps, I think that the argument is sufficiently straightforward. I would be interested in hearing if others results corroborate or contradict my argument.